Rare event simulation for probabilistic models of T-cell activation

(1)

Technische Fakult¨at

Biomathematik und Theoretische Bioinformatik

Rare event simulation for

probabilistic models of

T-cell activation

Der Technischen Fakult¨

at der Universit¨

at Bielefeld

vorgelegt zur Erlangung des akademischen Grades

Doktor der Naturwissenschaften

Dipl.-Inform. Florian Lipsmeier

Bielefeld, Germany, July 21, 2010

Biomathematics & Theoretical Bioinformatics Group

Faculty of Technology

Bielefeld University

(2)

ii

Supervisors:

Prof. Dr. Ellen Baake Prof. Dr. Sven Rahmann

(3)

iii

Acknowledgements

Undertaking and finishing a PhD project certainly requires great effort in many respects. Without help, support, and encouragement from several persons, I would never have been able to finish this work.

First of all, I would like to thank my supervisor Prof. Ellen Baake for many interesting and fruitful discussions, for keeping up with my sometimes confusing explanations and for allowing me much freedom in my research.

Thanks to all the members and former members of the Biomathematics and theoretical Bioinformatics group, especially for their patience with my many test talks during our group seminars and their helpful criticism.

Finally, I’d like to thank my family for encouraging me to pursue my studies and especially my companion in life Vera Drees for supporting me with your love and understanding.

I am grateful for the financial support I received from the NRW Graduate School for Bioinformatics and Genome Research.

Bielefeld, March 2010 Florian Lipsmeier

(4)

(5)

Nomenclature

¯

τ mean of the exponentially distributed mean

binding time between an antigen and a TCR

η relative error; standard deviation divided

by the mean

T random variable for the mean binding time

between an antigen and a TCR

τ binding time between an antigen and a TCR

ϑ tilting parameter

g density of W

G(z(f )₎ _{random variable for the total stimulus}

in-duced by all the antigens on an APC to a random T-cell, as a function of the copy number of the foreign antigen

gact activation threshold; value that has to be

reached by G(z(f )_{) in order for a T-cell to}

get activated

gthy thymic activation threshold; if this value is

reached during negative selection, the T-cell is induced to die

m(c) number of constitutive antigen types

pre-sented on an APC

m(v) _{number of variable antigen types}

presen-tend on an APC

W stimulation rate induced by a single antigen

to a random T-cell

w realisation of W

z(c) _{number of copies of an individual}

constitu-tive antigen type presented on an APC

z(f ) number of copies of the foreign antigen type

presented on an APC

z(v) _{number of copies of an individual variable}

antigen type presented on an APC

zs number of copies of a self antigen type

pre-sented on an APC in the new T-cell acti-vation model

AIRE autoimmune regulator; responsible for pGE

regulation

(8)

viii Contents

APC antigen presenting cell

BCR B-cell receptor

DC dendritic cell

IL interleukin; group of different second

mes-sengers

IS importance sampling

LD large deviation

MHC major histocompatibility complex

pGE promiscuous gene expression

pMHC peptide-MHC; complex of an antigen and

an MHC presented on an APC

SS simple sampling

TC1,TC2,TC17 different types of cytotoxic T-cells

TH1,TH2 different types of T helper cells

TCR T-cell receptor

TEC thymic epithelial cell

(9)

Chapter 1

Introduction

When it comes to T-cells in immunobiology a summary of the standard school book descriptions basically looks like as follows. There are cells of the innate immune system that directly recognise and attack a pathogen and there are the cells of the adaptive immune system that are able to learn to recognise and attack all the pathogens that circumvent the innate immune system. One important family of cells of the adaptive immune system are the so-called T-cells. These cells have a special receptor that helps them to detect molecules on cell-surfaces called antigens. When a pathogen enters the body, it leaves a trail of antigens which are recognised by one of the T-cells and this T-cell clears the body of the pathogen.

So, where is the problem one might think. This seems like a pretty straightforward description of a mechanism that is completely understood. However, as we will show, quite the opposite is true. It is possible to give some general, oversimplifying explanations but most of the rest is still very unclear. With this thesis we provide insights and new ideas to illuminate the T-cell’s ability to detect pathogens. Let us make clear why this is so important and justifies intense research.

If we observe ourselves and our environment by means of a microscope, we see that we constantly interact in many different ways with many different microorganisms at any given time. This can happen in a symbiotic way as it is with the bacteria in our colon that support digestion or with the bacteria on our skin. Often, however, it is a hostile interaction, that is bacteria, fungi or viruses try to invade our body to use it for their survival and for their reproduction. Thereby they can make us sick or even kill us, eventually.

Most people are living a healthy life, at least over long periods of their lifetime. We rarely get ill and if so, we normally get better quite fast. This is a quite amazing observation, though none of us thinks a lot about it. It is so amazing because our body is under constant attack from a huge variety of pathogenic microbes, like bacteria and viruses. A fact hardly acknowledged by anyone, as we only get to know an attack if it is successful. The counterpart to these microbes in our body is the so called immune system. This highly complex system of many different cells enables the recognition and killing of pathogens in most of the cases.

One central part of this immune system are the so-called T-cells. Since decades much research is devoted not only to T-cells but of course also to all other cells of our immune system. This led to new drugs and therapies in order to prevent diseases. Despite all these efforts much is still unknown.

T-cells belong to the part of the immune system that is termed adaptive because

(10)

2 Introduction

they have the ability to recognise unknown pathogens and in case of a second meeting with the same pathogen they are able to react much faster. T-cells can on the one hand support other cell types in the direct attack of a pathogen or on the other hand detect and destroy cells of our own body which are infiltrated by pathogens in order to reproduce. They are even able to detect mutated (cancer) cells to a certain degree and destroy them. Hence, it is highly advantageous to know all details regarding these cells. Surprisingly, there is still a central aspect with regard to T-cells that can only be explained insufficiently. T-cells are activated via the recognition of short amino acid sequences that are displayed on cell surfaces in the body. These sequences are residues from the degradation cycle of the cells and are called antigens. On this molecular level there is no definite characteristic that identifies an antigen as foreign, that is as coming from a pathogen, or as self, that is coming from our own cells. The T-cell constantly ’scans’ for antigens and mostly of course they encounter self antigens. A reaction to self antigens would lead to autoimmune reactions. A missing reaction to foreign antigens on the other hand can lead to severe diseases and death.

In order to explain the mystery of the recognition of a foreign antigen against a back-ground of many different self antigens, experiments only are not sufficient. Therefore, there are efforts to use the experimental knowledge as a basis for the development of mathematical models that explain foreign-self discrimination.

At this intersection of immunology, mathematics and computer science this thesis is situated. We use a combination of mathematical modeling on the basis of biological hypothesis and an efficient simulation method to explore the mechanism of T-cell acti-vation with emphasis on its foreign-self discrimination capability. We introduce some of the already existing models in the next chapter and then concentrate on one spe-cial model of T-cell activation developed by van den Berg, Rand and Burroughs (BRB model of T-cell activation)) in 2001 [205]. In contrast to the other models, this model describes T-cell activation probabilistically. We will explain why this is necessary and beneficial. Previous work could show that this model is capable of explaining foreign-self discrimination [205, 232]. This model is our starting point for the exploration of T-cell activation and foreign-antigen discrimination. From here we go on to extensions of the model and the development of a related model that captures the biological reality in a better way.

An integral part of this thesis is the development of a general and efficient simulation method for a certain type of stochastic models. We use this method for the analysis of the BRB model and its extensions. The development of a new simulation method is necessary because, as we pointed out, T-cells mainly meet self antigens and only very rarely foreign antigens. Hence, the recognition and activation of a T-cell has to be a rare event. But in order to investigate such rare events in detail, sophisticated methods are needed.

The other important part of this thesis is the development and analysis of a new T-cell activation model based on our results from the analysis of the BRB model and additional recent experimental findings. In this model we include a special ’educational’ mechanism during T-cell development termed ’negative selection’. Plainly speaking, this mechanism helps to sort out T-cells that are too self-reactive. We succeed in combining negative

(11)

3

selection with the central ideas from the BRB model and show its major influence on foreign-self discrimination.

Because of its importance we devote an additional chapter to a first modeling approach of T-cell migration in the thymus, the place where negative selection occurs. This opens up a new direction of research that should on the one hand clarify the scope of negative selection with regard to the foreign-self discrimination ability of T-cells. On the other hand it should be the first step to the better understanding of negative selection in order to find ways to actually manipulate this process to prevent autoimmune diseases and enhance the effectiveness of T-cells against certain pathogens.

In summary, our present work has three important cornerstones. We introduce all necessary biological details with regard to T-cells that not only suffice as a solid back-ground for this thesis but also as a starting point for future work. Furthermore, we develop a powerful simulation method whose application area goes beyond the models we explore here. Most importantly, we deliver new hypotheses and parameter estimates for T-cell activation, foreign-self discrimination and negative selection via the analysis of the models presented in this thesis. These results can be one step in the direction towards new experimental research in order to finally really explain T-cell activation and therewith foreign-self discrimination.

The thesis is composed as follows:

Chapter 2 − 4 deal with the development of the basic biological and mathemati-cal/informatics knowledge that is needed. In Chapter 2 we therefore introduce the immune system in general and concentrate then on T-cells. Afterwards we describe in Chapter 3 the BRB model of T-cell activation and finally in chapter 4 the basics for our simulation method.

Chapter 5 and 6 represent the core of this thesis. In chapter 5 we develop the sim-ulation method and prove its efficiency. Then we use it to analyse the BRB model. Furthermore we develop and analyse extensions of the model as a consequence of our first results. The sixth chapter is devoted to a new model of T-cell activation which we develop on the basis of our ideas gained in chapter 5. We describe the development of the necessary modified simulation method and test the model with different parameter values in order to explain foreign-self discrimination by T-cells.

Chapter 7 deals with the development of a new model of T-cell migration that should help to clarify the negative selection process. The model we develop in this chapter does not describe T-cell activation but as this mechanism has a vital influence on foreign-self discrimination of T-cells it fits well in the context of this thesis.

Finally we summarise our results in chapter 8 and give an outlook on the implications of our work on future research.

(12)

Chapter 2

Biological background

In this chapter we elaborate on the immunobiological background of this thesis. In this respect, it is our aim not only to introduce the necessary facts, but we incorporate them in a kind of review. Thereby we want to draw a picture of the newest relevant findings with regard to T-cells and present them in a way that leads to new approaches and ideas also beyond this thesis. We therefore decided to introduce special sections which summarise important data which we extrapolated from different publications as an asset for further model development.

This chapter consists of three parts. At first we give a short, very general introduction on the immune system in order to motivate the role of T-cells in this framework. The second part is concerned with the T-cell development process. This is necessary because it already conveys important information on the establishment of foreign-self discrimi-nation of the T-cell repertoire. Finally, we explore the mechanism of T-cell activation. Besides experimental and theoretical basics, we also introduce different models of T-cell activation that already exist.

2.1 The immune system at a glimpse

We and all kinds of (jawed) vertebrates share a similar complex defence system, called the immune system, which protects its host against all sorts of pathogenic microbes. Under the roof of the immune system there is a multi-faceted collection of cells, molecules and their interactions, which enable a specific and non-specific recognition and elimination of a variety of pathogens. Generally, we distinguish between two different parts of the immune system. There is the innate immune system, comprising all innate, that is non-specific, immune responses and there is the adaptive immune system, which comprises all specific immune responses. However, we have to point out that these two parts are entwined with each other in many different ways. Adaptive responses develop as innate responses occur. The innate immune system can therefore be seen as a first response unit, that reacts directly on a pathogen encounter. Furthermore, the adaptive immune system really is adaptive, that is, it allows for the development of a memory and thereby amplifies the reaction efficacy to new encounters with the same pathogen. Just in contrast, responses of the innate immune system do not vary, no matter how often the same pathogen is encountered.

(13)

2.1 The immune system at a glimpse 5 mucous membranes skin neutrophils, DCs stomach acid, enzymes

first line of defense

germs

DC macrophage

second line of defense third line of defense

inflammation

cytokines

Figure 2.1: The three lines of defence of the innate immune system. The first line tries to prevent the penetration of the body. It consists of a mixture of passive (e.g. skin, stomach acid, membranes) and active (e.g. special types of dendritic cells (DCs), neutrophils, macrophages) elements, where the former ones are just physical barriers and the latter ones are situated at/in this barriers and attack the possible intruder. The second line consists of different types of dendritic cells, neutrophils or macrophages which are situated inside the body but near the surface. Finally, the third line of defence produces an inflammatory environment via cytokines that facilitates DC and macrophage movement and their ability to attack germs.

2.1.1 Innate immune system

We are surrounded by myriads of microorganisms, some of which try to invade our body. To prevent this in the first place, we have different innate lines of defence (see Fig. 2.1) [152, 142]. The most obvious one is the skin and its inner-body equivalent, the mucous membranes (inner-body surfaces which are exposed to the environment). Both form a physical barrier through so called tight junctions, which are firm cell-cell connections. They are furthermore covered with different epithelial cells. Those on our skin form a dry, protecting layer which is hardly penetrable. This is why most of the pathogens try to overcome the mucous membranes instead. Their epithelial cells utilise different mechanisms to prevent this intrusion. They produce mucus to trap microbes, propel them away using cilia and they produce special enzymes and anti-bacterial peptides to eliminate pathogens. Additionally, they facilitate the colonisation of their surface with friendly microorganisms which compete against foreign microbes and can synthesise anti-bacterial substances.

If this first barrier is penetrated, for example by an injury of our skin, the second barrier comes into play. It comprises special cell types which are capable of endo-or phagocytosis, that is the internalisation and degradation of either macromolecules (endocytosis) or whole microorganisms (phagocytosis) [105, 200]. The main cell types responsible for these actions are the dendritic cells (DCs), macrophages and the neu-trophils. They identify targets via pattern recognition receptors on their cell surface that bind to surface molecules of invading microorganisms [132, 101].

Before proceeding with the innate immune response it is at this point necessary to have a closer look on receptors and receptor binding, as this is crucial not only here, but also for the adaptive immune system and in particular for this thesis. In order to ’see’ their

(14)

en-6 Biological background

vironment all cell types rely on different types of receptors. These are special molecules which are embedded into the membrane of a cell. Mostly they are trans-membrane molecules that detect molecules with their extracellular part and start signalling with their intracellular part. Depending on aspects as different as three-dimensional con-formation, charging, amino acid composition and so on, different molecules (normally called ligands, in the special case of immunology we speak of antigens) bind with dif-ferent affinities to a receptor. If a molecule with a certain affinity binds to a receptor, the receptor is activated and starts its task, which normally involves the start of some type of intracellular signalling cascade. There are quite unspecific receptors which re-act to many ligands with low binding affinities, as well as very specialised receptors, which only react to one or two well-fitting ligands with very high binding affinities. In general ligands are peptides, hormones, toxins or drugs. In the special case of foreign microbe recognition by T-cells, they are normally just peptide chains (strings of amino acids) which result from a preprocessing of the native (three-dimensional) antigen and are presented by a special class of cells, called antigen presenting cells. As there are 22 different amino acids which can be concatenated to peptide chains in all combinations, the number of possible antigens is very high, which will come into play later. Germ detection by other cell types of the immune system is not so restricted. Their receptors are able to detect antigens in their native, three-dimensional form.

Receptors used in the innate immune response are germline encoded and therefore cannot be changed. That is, every receptor type is hardcoded by a specific gene. This is why there is only a limited number of receptors available in a host. To overcome the disadvantage of this restricted repertoire in the face of the enormous variety of pathogens, these receptors are quite unspecific (also termed crossreactive) concerning the targets they bind (e.g. CD14 is a receptor which binds to all kinds of bacterial lipopolysaccharides). They detect special patterns that are pathogen associated [132]. Thereby they can detect whole families of pathogens that share structural elements which are detected by such a pattern recognition receptor.

A third barrier of the innate immune system is formed by inflammations, which come about if tissue is damaged or if pathogens are recognised [142]. Inflammatory reactions facilitate the movement of effector molecules and cells to the affected area in order to support the killing of the pathogens. Furthermore, the infection is restricted through the healing of the tissue and the building of physical barriers to prevent a further movement of the pathogens.

It is obvious that, although fast and highly effective in killing pathogens, the innate immune system has its Achilles heel in the restricted number of possible receptors for the recognition of pathogenic organisms. In the course of evolution microbes have developed numerous ways to prevent recognition, e.g. by the development of a thick polysaccharide capsule. Even if they are recognised some microbes are capable of (mis-)using the degra-dation cycle of macrophages for their own purpose and grow within these macrophages. The answer to these threats was the development of a second detection system, which had to be more specific in its recognition mechanism and not restricted to a limited number of receptor types.

(15)

2.1 The immune system at a glimpse 7

2.1.2 The adaptive immune system

It is an inherent feature of evolution that through mutation and selection it produces either new variants of a species or even new species. The rate of change of a species is dependent on the length of its reproduction cycle, as during reproduction the important changes occur via genetic mutation and recombination. Most of the microorganisms have very short reproduction cycles, especially compared to vertebrates. Hence, they can change faster and thereby avoid detection. It follows that to prevent bypassing of the host’s immune system, there were two possible ways in evolution to go. The first one would have been the expansion of the repertoire of germline encoded receptors. This happended probably during the evolution of invertebrates, which often have a much bigger repertoire of these receptors than vertebrates [151]. The second way has been the development of a new detection mechanism, which is more flexible and specific in its pathogen detection capability. This came about during the evolution of (jawed) vertebrates. Although no one can say for sure whether this way was the optimal one (in terms of selective pressure during evolution), there exists a convincing hypothesis. In vertebrates there is a huge variety of microorganisms, living in symbiosis with the host. In order to guarantee the safety of these symbiotic arrangements, the vertebrates had to deplete their innate receptor repertoire of all receptors which could recognise antigens from these ’friendly’ microorganisms. This would have made the host also more vulnerable to other pathogenic microorganisms [151, 163]. The development of a second, adaptive, immune system with a receptor repertoire that is more specific was therefore necessary. This section highlights the most important aspects of this adaptive immune system, with special emphasis on T-cells.

The adaptive immune system has a complex task. It should defend its host against all pathogenic microbes that circumvent the innate immune system and thereby deal with all occurring new mutants of such a microbe, keep it then in a kind of memory to act faster on the occasion of a second infection attempt and of course it should not attack the host itself or microorganisms living in symbiosis with its host. These are exactly the very characteristics of the adaptive immune system [142, 152]. There are two key players involved in adaptive immunity, the so-called lymphocytes and the antigen-presenting cells (APC). Both can enter lymphoid organs and tissues and otherwise circulate around the body by means of the vascular and lymphoid system.

Antigen presenting cells are, as the name tells, highly specialised cells, which can internalise and present antigens. The most important group of APCs are the dendritic cells (see Figure 2.2). Here, we can see one of the several connections between the innate and the adaptive immune system. DCs also play a role in the former one because they internalise pathogenic material, as already mentioned. However, they have developed a very specialised signalling and receptor apparatus that makes them the ideal interaction partner for one group of lymphocytes, so-called T-cells. Another group of APCs are special lymphocytes called B-cells, which besides being APCs serve other important purposes, which is explained in the next paragraph. An APC produces so called major histocompatibility complex molecules (MHC) and acquires antigens. An MHC needs the antigen to form a stable macromolecule, which is then expressed on the surface of an

(16)

8 Biological background

APC as an antigen (or peptide)-MHC (pMHC) complex. An antigen can only bind to some of the several types of MHC molecules. Currently, for humans 3371 different MHC alleles are known which belong to two different classes (2351 for MHC class I, 1020 for MHC class II) (see www.ebiac.uk/ipd for updated numbers), from which only very few are expressed in every single individual [160]. In fact all cells in the body express MHC I molecules on their surface together with fragments out of their interior, but APCs are especially equipped to produce them in great numbers together with many more antigens.

The pMHCs on the APCs are scanned by the T-cells.

Anti-Figure 2.2: A dendritic cell equipped with antigens

gen acquisition is primarily done by the already mentioned mechanisms of phago/endocytosis. The internalised cell ments are degraded and the resulting very small peptide frag-ments form the antigens. Figure 2.3 illuminates this, for the present thesis, very important mechanism. The different as-pects of it are explained in detail in different sections of this chapter.

At this point it is important to note that an APC has only very limited capabilities of pathogen detection through pattern recognition receptors. In principle it internalises all sorts of cell material from its surroundings. This implies that most of them are parts of dead cells from the host itself. Hence, most of the antigen produced fall into the category of so-called self antigens, that is antigens of the host itself. On the molecular level there exists no distinction between a foreign and a self antigen, that is they are just strings of amino acids with no special marker for pathogenic material. As long as no infection occurs we have to assume that there are even no pathogenic antigens presented by an APC. However, these APC also lack an activation signal that enables them to stimulate T-cells. This signal is supplied via the pattern recognition receptors if they detect pathogens. A signal of these receptors does not only activate the APC, but also leads to an enhanced incorporation of (presumably) foreign antigen [150, 132]. However, it should be clear that even APCs that encounter pathogens will mostly present self antigens because they are flooded with them constantly. This is a crucial observation with regard to the topic of this thesis. The recognition of the limited amount of foreign antigens against a background of many self antigens. A reaction to self antigens would lead to autoimmune reactions and a missing reaction to foreign antigens would let the pathogen invade the body. Furthermore, APCs can co-determine the reaction of the T-cells to the pMHC by means of other types of molecules such as co-receptors or cytokines, which will be elucidated later.

Lymphocytes constitute 20-40% of the white blood cell population. Their two major subsets are the already mentioned T- and B-cells.

As this thesis deals with T-cells, we only briefly outline the function of B-cells. They are responsible for the humoral immune response, that is they induce a reaction on sub-stances in the extracellular fluids [142]. B-cells are generated in the bone marrow, where they also mature. During maturation they develop a special receptor, the B-cell receptor (BCR), which is unique for every B-cell. This is achieved by the genetic rearrangement

(17)

2.1 The immune system at a glimpse 9 APC TCR T cell MHC molecule various peptides internalise digest

Figure 2.3: A T-cell and an antigen-presenting cell (based on Fig. 1 of [206]). An APC absorbs molecules and particles from its vicinity and breaks them down. The emerging fragments, so-called peptides (short sequences of amino acids), serve as antigens. They are bound to so-called MHC molecules (still within the cell), and the resulting complexes, each composed of an MHC molecule and a peptide, are displayed on the surface of the cell (the MHC molecules serve as “carriers” or “anchors” to the cell surface). Since most of the molecules in the vicinity of an APC are “self” molecules, every APC displays a large variety of different types of self antigens and, possibly, one (or a small number of) foreign types. The various antigen types occur in various copy numbers. Each T-cell is characterised by a specific type of T-cell receptor (TCR), which is displayed in many identical copies on the surface of the particular T-cell. When a T-cell meets an APC, the contact between them is established by a temporary bond between the cells, in which the TCRs and the MHC-peptide complexes interact with each other, which results in stimuli to the T-cell body. If the added stimulation rate is above a given threshold, the T-cell is activated to reproduce, and the resulting clones of T-cells will initiate an immune reaction against the intruder.

of the genes responsible for the expression of this receptor (VDJ recombination) [126]. A very similar mechanism is also involved in T-cell development, therefore we skip any details. With the BCR a B-cell can bind to free (soluble) antigens in their native form. To prevent autoimmune reactions to self antigen, all B-cells have to undergo a selection process during maturation. They meet a huge amount of self antigens in a special envi-ronment and only survive if they do not react [2, 144]. On the encounter of its cognate (= perfectly fitting = agonistic) antigen outside the bone marrow, the B-cell is activated and can become either a plasma B-cell, which secrets antibodies (exact copies of its BCR), which bind to every cognate antigen and thereby mark the associated microbe for termination, or it can become a memory B-cell, which lives for a long time and can react much faster to a second encounter with the same antigen (see Figure 2.4). So B-cells fulfill all the requirements mentioned above for the adaptive immune system.

T-cells

T-cells belong to the group of white blood cells (lymphocytes). They develop in the thymus and form an integral part of the adaptive immune system, with plenty of different tasks. Mainly they detect and attack pathogens which bypass the detection via B cells

(18)

10 Biological background Germs B cell B cell memory B cell memory B cell antibodies

Figure 2.4: B lymphocytes are responsible for the humoral immunity. That is, they detect soluble antigens (small germs or toxins) in the blood via their special B-cell receptor (BCR). This BCR binds to antigens in their native, 3 dimensional shape (in contrast to the T-cell receptor as we will explain later). After a successful detection, the B-cell can release antibodies, which are exact replicas of their BCR. These antibodies bind to all cognate molecules and surface proteins and thereby mark them for other cells like macrophages for termination by phago/endocytosis. Some of the activated B-cells become memory B-cells that survive in a resting-state and can react much faster on a second encounter with the same antigen.

and their antibodies. Moreover, they support B-cells with the effect of new cognate antibody production. One subpopulation of T-cells, called regulatory T-cells (Treg), even acts to stop immune responses in order to prevent autoimmune reactions [142].

A T-cell carries several copies of a specific unique receptor, called T-cell receptor (TCR), on its surface. Upon leaving the thymus, T-cells migrate through blood vessels and especially the lymph nodes. If they encounter an APC, they sort of ’scan’ it via their TCR, that is the copies of their TCR bind to the pMHC molecules on the surface of the APC. Every TCR can only bind to one (or a very restricted set) of the several MHC types, which is the first step in a stable binding of the TCR-pMHC complex. The second step is the binding to the presented antigen. If this binding is stable enough, that is the binding duration of such a complex exceeds a certain threshold, the TCR is triggered, signals this to the T-cell and eventually the T-cell is activated.

Generally, naive T-cells, that is T-cells which have not been activated, belong to either of two main types: CD8+ CD4− (cytotoxic) and CD4+ CD8− (helper) T-cells (the + in CD4+ just means that this receptor is expressed on the cell surface. A − means the opposite. In the following we omit the + and the whole − term). These two molecules are important co-receptors besides the regular T-cell receptor. CD8 helps to stabilise the binding of its T-cell with a MHC molecule of class I, whereas CD4 helps with the binding to MHC II molecules. Tregs are a special group of CD4 T-cells with additional

(19)

2.1 The immune system at a glimpse 11

CD25 molecules. Less than 10% of the T-cells have neither of the two molecules. These special T-cells do not bind to pMHC complexes, but have different other molecules which enhance the binding to for example glycolipid antigens [37].

T-cell differentiation

Without going into the details of T-cell activation, which forms the integral part of this thesis and will be described in greater detail later on (see Section 2.3), let us assume that a T-cell is activated by a pMHC complex on an APC. Both, activated CD4 and CD8 T-cells start a rapid proliferation process in order to generate many clones. Furthermore, the activation event leads to a differentiation of the T-cell clones into various possible T-cell types, depending on the co-stimulatory signals and cytokines being present during the activation process [141]. Until now, 4 main subpopulations of helper T-cells and 3 main subpopulations of cytotoxic T-cells were identified. In the presence of interleukin 12 (IL-12) naive CD4 T-cells become TH1 cells, which support cell-mediated immune

responses by secreting the second messengers IL-2, IFNγ and TNF-α and thereby com-municate to other cells. This leads, for example, to a an improved killing efficacy of macrophages and an improved proliferation of cytotoxic T-cells as well as production of antibodies. They also support TH1 differentiation through a positive feedback loop

[191, 180, 122].

Activation of naive CD4 T-cells via IL-4 stimulated dendritic cells lead to a differen-tiation to TH2 cells [51], which support humoral, antihelminthic (against worms) and

allergic immune responses [227, 44]. By the production of 4, 5, 6, 10 and IL-13 and GATA-3, they enhance B-cell proliferation and antibody secretion. Additionally, they inhibit TH1 cell differentiation while simultaneously promoting TH2 differentiation

[223].

A third very recently found subpopulation of helper T-cells are the TH17 cells, which

are developed in the presence of TGF-β and IL-6. They owe their name to the cytocine IL-17, which they produce beside IL-17a and IL-22. Due to the broad distribution of IL-17 and IL-22 receptors, they thereby induce a massive tissue reaction. Consequently, they promote tissue inflammation (especially during autoimmune diseases). A second very important task of TH17 cells is the clearance of extracellular pathogens during

infections [188, 109, 147].

A special group of CD4 T-cells are the regulatory T-cells . For these cells two distinct ways of development were identified [38]. A larger part of the Treg cells are CD4 CD25 natural Tregs, which are indispensable for the maintenance of immunological tolerance in the host, that is they prevent autoimmune reactions [171, 170, 172, 149]. These Tregs are thymus-derived, similar to the naive CD4 and CD8 T-cells. In contrast to these, the second group, so-called induced Tregs, are generated during CD4 T-cell differentiation upon activation and the simultaneous presence of IL-2 and TGF-Beta. This is important for achieving oral tolerance to allergens and other antigens. It has been found that differentiation of these Tregs and TH17 cells is reciprocally regulated [226]. All these

(20)

Figure 2.5: CD4 T-cell differentiation; During activation, the T-cell receives different stimulation and inhibition signals in form of different interleukin molecules. These signals influence the differentiation of the T-cell into different types of activated T-cells, which then also fulfill different tasks in the immune reaction. Consequently a naive CD4 T-cell can ultimately influence the immune reaction in all different parts of the immune system, that is humoral, cellular and innate immunity.

In a similar manner CD8 T-cells differentiate into the subpopulations TC1,TC2, and

TC17, whereby these play a much more active role in comparison to T helper cells.

[223, 120, 137]. After activation they scan all cells in their surrounding and as these cells also present MHC I molecules with antigens (but only in small numbers), they might encounter their activating antigen again if the pathogen is reproducing inside the cell. If so, they attack this cell and thereby prevent this reproduction.

All in all, it is important to note that the whole process of T-cell activation and differentiation is much more than a binary on-off decision upon antigen encounter. On the contrary, the final outcome is very much dependent on a well-orchestrated mixture of different second messengers that bind to other receptors on the T-cell. However, we are only interested in the very first step, namely the activation signal induced by antigens and we assume that this process follows the same rules for all types of T-cells.

2.2 T-Lymphopoiesis

After the short general introduction to the immune system and T-cells, we now focus on all the different aspects regarding T-cells, starting in this section with their developmen-tal process (T-Lymphopoiesis). We will explain why this can already help to understand many aspects of foreign-self discrimination.

T-cells as well as the other lymphocytes (B-cells , natural killer (NK) cells), but also a portion of dendritic cells, develop from the same origin, so call hematopoietic stem cells (HSCs) [19]. These cells are situated in the bone marrow and are the source of all types of blood cells. Through a complicated signalling network, HSCs differentiate

(21)

2.2 T-Lymphopoiesis 13

into different progenitor populations, ultimately resulting in all lineages of blood cells. For most of these cell types, the maturation process takes place in the marrow, with one important exception, the T-cells. For their development a whole organ, the thymus, stands by [41].

At the top of the T-cell development hierarchy in the thymus are the so called thymus-resident T-cell progenitors. These cells originate from the bone marrow and are period-ically imported into the thymus [66]. The continual settling of new progenitor cells is necessary due to the fact that these cells have only a limited self-renewing capacity. The wave-like behaviour of the progenitor cell influx into the thymus accompanied by a wave of intrathymic DC formation is assumed to optimise T-cell selection, the ultimate stage of T-cell development in the thymus [60]. T-cell selection helps to sort out inactive or autoaggressive T-cells and is described in detail later.

During T-Lymphopoiesis three main stages are identified. They are called double neg-ative (DN), double positive (DP) and single positive (SP), depending on the expression of none of the co-receptors CD4 and CD8, the expression of both of them and ultimately the expression of only one of the two co-receptors [183]. This differentiation leads to two different T-cell subpopulations, whose functions have already been characterised in Section 2.1.2. Each of the three stages occurs in different areas of the thymus consisting of different microenvironments [5]. Actually, it is important to note that the develop-ment of thymocytes is crucially dependent on the interaction with thymic epithelial cells (TECs) in the different microenvironments, but on the other hand TECs need the in-teraction with thymocytes to develop the appropriate microenvironments [210, 209, 21]. Thymocyte development and migration through the thymus seems to be governed by the sequential expression of different chemokine receptors and the release of chemokines via TECs in individual microenvironments [192].

The first two stages occur in the cortex of the thymus. During the DN stage the lymphocytes migrate to the outer cortex called subcapsular zone. Meanwhile they pro-mote the development of cortical TECs (cTECS) from TEC progenitors and start the development of the already mentioned T-cell receptor (TCR). This receptor consists of two different protein chains. The successful assembly of the first protein chain and the formation of a pre-TCR complex on the cell surface marks the transfer of a T-cell from the DN to the DP stage. In this second stage T-cells fully develop the TCR and express it in low levels on its surface besides the two co-receptors CD4 and CD8. They now undergo the process of positive selection, during which most of the T-cells die [63, 73]. After surviving positive selection, T-cells switch to the SP stage and migrate to the medulla. Here they encounter a negative selection mechanism, which can beside killing a T-cell, also turn it into a Treg cell. The few surviving T-cells finally enter the periphery in order to defend the host. In the following we highlight some of the important aspects of T-cell development, including these selection mechanisms, to clarify this on the first sight rather complicated developmental process.

As mentioned briefly in the introduction, T-cell activation is dependent on the TCR, which will be highlighted in Section 2.3. It is therefore necessary to have a closer look at this T-cell receptor. Indeed, the TCR of a T-cell will be its defining element throughout this thesis.

(22)

Figure 2.6: T-lymphopoiesis. T-cell precursors enter the thymus and start to interact with the thymic microenvironment. In the 3 double negative stages (DN1, DN2, DN3) they express no co-receptor and start to develop the β-chain of the TCR. The successful assembly of a β and pre-α-chain marks the transfer to the double positive (DP) stage where both co-receptors, CD4 and CD8, are expressed. The T-cells undergo positive selection during which different α-chains of the TCR are generated until the T-cell receives signals from the cTECs or dies by neglect. Ultimately, the T-cells move from the thymic cortex to the thymic medulla and by downregulation of one of the co-receptors enter the single positive stage (SP). Then the T-cells meet with mTECs and DCs which can induce death by apoptosis. Finally, after circulating around for some days in the medulla, the surviving T-cells leave the thymus.

During the DN stage lymphocytes start to generate the so called T-cell receptor (TCR). Generally, two different classes of TCRs are identified, the αβ and the γδ recep-tor. T-cell progenitors become committed to one of these two different T-cell lineages [20].

2.2.1 The T-cell receptor

A T-cell receptor is a heterodimer that consists mainly of two transmembrane glycopro-tein chains, α and β or γ and δ. Each of the proglycopro-tein chains is anchored through the cell membrane and consists of two extracellular domains, a constant one and a variable one. It is this variable domain that makes the TCR so special and allows for the detection of many different antigens. The domain is called variable, because in the process of its generation the expression of the associated genes undergoes a procedure called V(D)J recombination [4]. This procedure rearranges variable (V), diversity (D) and joining (J)

(23)

gene segments, such that every variable domain and thereby every TCR becomes almost unique. This is true not only for every single individual, but also, to a certain degree, in between different individuals. Recently is has been shown that there are certain TCRs which are shared by groups of individuals. This is due to “convergent recombination”, that is their probability to be expressed is higher because they can be built up from several genetic combinations (degenerated genetic code, alternative splicing possibili-ties), rather than because of a bias in recombination [213]. Furthermore, the full TCR complex consists of several other protein chains, serving as co-receptors and having a direct influence on the downstream signalling events inside the T-cell. Figure 2.7 shows a cartoon version of the full TCR complex.

The generation of the TCR follows a strict

or-α β Ag ε γ γ

_ε

CD3 CD3 CD4 I T A M variable segments CD3 ζ

Figure 2.7: TCR complex; the α and β chain are the actual TCR, whereas the other protein chains play the role of co-receptors. ITAMs (immunoreceptor tyrosine-based activation motif) are seg-ments of the CD3ζ chains that are phos-phorylated during TCR-antigen binding by receptor-associated kinases like LCK. This phosphorylation is the starting point of T-cell signalling and is also influenced by the co-receptors.

der in which the genes for the different protein chains are recombined sequentially (δ > γ > β > α) [59]. Interestingly, γδ T-cells can generate more unique receptors than αβ T-cells and B-cells combined, but their repertoire is quite re-stricted only to a specific subset of these recep-tors [27]. The role of γδ T-cells in the immune system is not well understood until now. They play a role in innate as well as adaptive immune responses and have different functions depending on the tissue they reside in. Indeed, there seems even to be a particular association between cer-tain γδTCR repertoires and cercer-tain tissues [84]. Receptor binding and activation is not MHC re-stricted and they are assumed to bind to so-called antigens, that is non-peptidic, phospho-rylated compounds. It is assumed that they, for example, influence DC functions and thereby their immune responses to infectious pathogens and/or are a complement and regulator of αβ T-cells [29]. γδ T-cell activation seems to rely mainly on self antigens which are expressed by stressed or (near)-apoptotic cells, due to for example tumor development or infection by viruses [195], just in contrast to αβ T-cells which are strongly dependent on foreign antigens and have all the different functions already described previously in Section 2.1.2. Therefore, αβ and γδ T-cells are two very dis-tinct T-cell lineages which have not very much in common and need to be examined separately. In the following we omit the γδ T-cells and concentrate on αβ T-cells, the integral part of the adaptive immune system.

As mentioned earlier, the checkpoint for a T-cell to turn from DN to DP stage is the development of a TCR complex. This checkpoint is called β-selection, as this pre-TCR complex contains the successfully assembled β chain, together with a pre-α chain. The latter chain is then also assembled and the T-cell has a fully functional TCR. From this point on, the survival of a T-cell is crucially dependent on its receptor. A T cell

(24)

regularly needs certain signals in order to survive. These signals can only be induced by the binding of its TCR to peptide-MHC complexes on the cell surface of antigen presenting cells. This is true for the thymus, where the TECs and DCs play the role of APCs, as well as for the periphery where there are further types of APCs. As a result of this dependence, T-cells start to scan cTECs and DCs in the cortex for pMHCs and if their receptors are not able to bind to any of these pMHC complexes, the T-cells die (death by neglect). This process is called positive selection.

2.2.2 Positive selection

T-cell receptors as such are just macromolecules with the capability to bind to cell surface proteins. In this respect they are not different from for example antibodies. This also implies that initially there should be no bias for a T-cell to bind to pMHC molecules on specialised APCs, but given the broad spectrum of possible surface proteins in an individual, it is very probable that every freshly produced TCR could bind to several of these surface proteins [69]. On the other hand T-cells need to specifically bind with their TCRs to pMHC complexes on APCs in order to receive surviving signals and ultimately detect foreign antigens. Consequently, not every T-cell produced in the thymus should be able to fulfill these task, but every T-cell circulating in the periphery has the ability to bind to one type of pMHC complex and thereby to receive surviving signals. Furthermore, it is obvious that there cannot be a common binding site for all types of MHC molecules that can be recognised by every TCR, because then every T-cell in the periphery would recognise all MHC types. The specificity of the TCR suggests that the binding site of the TCR to the MHC is built either completely randomly during V(D)J recombination or the genetic combinations are restricted such that a TCR can bind to only one MHC type. In either case, many T-cells are produced which cannot bind to one of the very few MHC types present in an individual. It is obvious that these T-cells are superfluous and further maturation of them would be a waste of energy for the thymus. Therefore it is necessary to sort out these useless cells, which is fortunately quite easy. The thymus itself has to take no active role in it, but acts as a passive bystander. It is this passivity, which eventually leads to the death of anergic T-cells. Up to this point developing T-cells have constantly received signals from TECs to ensure their survival and development. With a fully functional TCR developed, T-cells rely on a new surviving signal, the activation signal of the TCR, which is induced by a sufficiently long-lasting binding of the TCR to a pMHC on an APC. T-cells which do not receive such a signal die (death by neglect). This process is termed positive selection [184]. During the whole phase between successful β-selection and either successful positive selection or cell death, the T-cell is able to change the TCR α-chain via rearrangements several times [153] and thereby enhance its chance to get a positive selection signal. T-cells in the DP stage survive about 3-4 days, which limits the number of possible rearrangements. It is assumed that this restricted time window is necessary to regulate the TCRα repertoire [79].

(25)

Antigens stabilise the binding of the TCR with a pMHC complex. Therefore it is obvious that, even if the threshold (in terms of TCR-pMHC half-life) for TCR triggering (and thereby positive selection) is low, antigens play a role in it. Several experiments have shown that, at least under certain conditions, there is no need for a diverse self antigen repertoire in the cortex in order to induce positive selection [95, 96]. In fact, only a single antigen is needed to ensure positive selection of quite a diverse T-cell repertoire, which hints to a very low threshold for TCR triggering. Recently it could be shown that cTECs have a unique complex for proteolysis (the degradation of proteins), termed the thymoproteasome [139]. It belongs to the same family of proteasomes that are responsible for the generation of antigenic peptides presented by MHC I molecules in the periphery. The thymoproteasome fulfills the very same task in the cTECs. In contrast to the other proteasomes, it has a weak chymotryptic activity (chymotrypsin is the enzyme responsible for the proteolysis), such that only a unique restricted repertoire of MHC I associated antigens can be generated [193]. It is further hypothesized that for MHC II associated antigens a similar protein cleavage process occurs, involving cathepsin L or S, which is highly expressed in cTECs but not in mTECs [12, 90]. Consequently, positive selection seems to depend on the recognition of MHC in conjunction with a special restricted set of antigens, which are generated by the so called ’modest’ cTEC protein degradation [193, 140].

For a long time, positive selection has been seen to be the process that shapes the T-cell repertoire such that every T-cell binds uniquely to a specific MHC. But research into the reasons for organ rejection after transplantation has already shown that T-cells of the host were able to bind to foreign pMHC molecules of the organ, which implies a certain crossreactivity of a TCR to several MHC molecules. In several experiments the T-cell repertoire before positive selection was tested for its crossreactivity and it could be shown that many of them are crossreactive even to the few MHC molecules of the host itself [230, 69]. The same is true for the repertoire after positive selection. In the beginning there is actually even no restriction on either of the two MHC classes [65]. These observations are clarified by the investigation of TCR-pMHC crystal structures. Although the part of the receptor which binds to MHC molecules is generated randomly, there seem to be restrictions such that this randomness is almost limited to a specific pool of sequences. On the other hand an MHC molecule has several surface residues to which a TCR can bind, so called ’codons’ [69]. Thus, there is no shared structural binding epitope for all MHC molecules, but a collection. Both, this collection of codons and the pool of TCR binding sequences may have co-evolved, such that many TCRs are in fact quite crossreactive with respect to MHC molecules. All these observations imply that the role of positive selection in rejecting T-cells may be overrated [65].

Its role in the CD4 CD8 lineage decision is still not clear. The strength of the signal induced by pMHC complexes may play a role in it [82, 177]. Here the restricted repertoire of selecting antigens may play a role. There appears to be a negative/positive feedback from TCR-pMHC I/II complexes which influences this decision [65].

Nevertheless it is a fact that T-cells in the periphery do not bind to different MHC molecules, hence, are restricted to one class of MHC. Furthermore, T-cells also rarely bind and react to pMHC complexes where the peptide is a self antigen. Therefore, there

(26)

has to be another process which shapes the T-cell repertoire that leaves the thymus. This process is the last checkpoint a T-cell has to pass before entering the periphery and is termed negative selection.

2.2.3 Negative selection

During the DP stage, the TCR is not fully up-regulated. Hence, the TCR sensitivity is biased toward low-affinity interactions. Upon positive selection T-cells reach the SP stage. They are now committed to either the CD4 or CD8 lineage and express a fully functional signalling apparatus, such that in principle they could be released into the periphery in order to meet with APCs and look for pMHC molecules with high affinities to their TCRs in order to get activated [45]. Unfortunately, most of these presented peptides are self antigens, that is antigens derived from proteins of the host itself. A reaction to such antigens would lead to an autoimmune reaction and eventually the death of the host. This is obviously not the usual case in reality. Furthermore, it remains to be clarified why T-cells are restricted to one type of MHC and what the advantage of this restriction is. All of this can be explained by a close look at the last selection step in the thymus, negative selection. In short words, during this process all T-cells which react too strongly to self pMHC are killed and how this is achieved will be the topic of this section.

In the SP stage T-cells migrate to the medulla. During this migration and later in the medulla they constantly meet APCs in form of DCs, macrophages, cTECs and mTECs [192, 129]. These cells present self-derived pMHCs and many of the T-cells encounter their cognate antigen while scanning these cells, which leads to strong signals induced through the triggered TCRs to their T-cells. In the periphery this would lead to a T-cell activation, but in the controlled environment of the thymus the opposite is happening. A T-cell which is triggered too much induces its own apoptosis program. Although, in principle, in the cortex T-cell triggering (for full activation, in contrast to ’just’ receiving survival signals) and deletion are also possible, it is far more unlikely [87, 116, 129]. Besides TCR-pMHC binding duration, T-cell activation is also dependent on extrinsic factors such as co-stimulatory molecules. Only in the medulla these are expressed in sufficient amounts (comparable to the periphery) by mTECs and DCs, such that strong TCR triggering is much more likely.

It is this process which also imposes the strong MHC restriction on the peripheral T-cell repertoire [221, 95, 110]. All T-cells that have the ability to bind to several of the few different MHC types present in a host have a huge disadvantage during negative selection. As already mentioned before, different MHC molecules can present different antigens. Hence, if a T-cell can bind to several MHCs the probability to encounter a cognate antigen is elevated significantly and it is very improbable that this T-cell survives.

(27)

The need for MHC restriction and diversity

At this point it is appropriate to think about the reasons behind the MHC restriction and the reason why there are different MHC types available. MHC restriction leads to a T-cell reaction in a controlled environment. Without the restriction to one MHC type, the probability of activation by self antigens would be quite high, especially if there would be no restriction to MHC at all, which would be thinkable, as TCRs can bind to several other molecules. On the other side it is advantageous to have more than one MHC molecule type in a host, in order to counter evading strategies of pathogens. By means of mutations and recombinations these change their peptides in the course of evolution, which could lead to antigens that cannot be connected to the given MHC molecule and thereby detection via T-cells would be impossible. Thus, a diverse repertoire of MHC types is necessary. However, as mentioned before, for a single individual this diversity is very restricted, whereas over the whole population it is very high [160]. There is the paradox of high inter-individual diversity and low intra-individual diversity [221]. This can be explained best out of the perspective of evolution and selective pressure. If a single individual has too many different MHC types many more T-cells would be depleted from the repertoire during negative selection, because many more self antigens could be presented. Moreover, the risk of an autoreactive T cell to escape negative selection rises, as it is impossible to present all self antigens to a T-cell. On the other hand, this puts a single individual at risk to be defenceless to a mutated pathogen. However, it is very unlikely that the same pathogen can circumvent the immune system of other individuals of the same populations, because they have many distinct MHC molecules. Establishing a diverse repertoire of MHCs in an individual is therefore restricted by the risk of autoimmunity, while it is enhanced over the total population to ensure the survival of the other individuals [221].

How to obtain self antigens

Up to this point we have been generally concerned with self antigens that are presented via APCs in the thymus, but we ignored a crucial question. Where do these self antigens actually come from? In the periphery, APCs constantly collect cells, cell fragments, proteins etc., digest them and thereby produce antigens. It is thinkable that this is also true for the thymus. However, we have to think about the thymus as a very special organ, with very special tasks, that are different from all other tissues in the body. Therefore, the proteins and cells involved might also be different from those in the rest of the body and thus also the resulting antigens. This could be true for regular antigens, resulting from molecules involved in the normal cell cycle of every cell, and is definitely true for tissue restricted antigens (TRAs), degradation products of molecules which are only present in a certain tissue. Furthermore the body changes while developing and new kinds of cells and proteins occur. If the T-cell repertoire is not prepared for these, autoimmune reactions are provoked.

The thymus is often described as an autarkic regime, which releases T-cells, NK cells and DCs, but lets nothing inside from the periphery. This is not true at all. In fact

(28)

dendritic cells constantly migrate from the periphery to the thymus, loaded with pMHCs [94]. Recently it could be shown that this is true for a specific group of DCs, termed circulatory DCs, whereas there are also groups of DCs that are thymus residents [155]. Until now it is not clear how both types of DCs are involved in mediating central toler-ance. For thymic DCs it could be shown in vitro that they mediate Treg development [217]. Circulatory DCs seem to play a role in both negative selection and Treg induction [155]. It is clear that this constant influx of pMHC can mediate autoagressive T-cell dele-tion at least for the regular antigens, although it is not clear how the thymus manages to prevent DCs with foreign pMHC to enter the thymus, which would be devastating for immune protection. Very recently it could also be shown that there are also certain T-cells that migrate back to the thymus [80, 23]. Speculations on their role in the thymus are even more diverse, from the maintenance of certain thymic microenvironments, over import of self pMHCs up to deletion of autoreactive T-cells or the conversion to Tregs.

While circulatory DCs might carry enough pMHCs to mediate tolerance to regular antigens, it is improbable that they present enough tissue restricted antigens, let alone antigens from proteins involved in later stages of host development. Therefore, there is the need for other tolerance mechanisms. Until recently it was proposed that periph-eral tolerance (that is tolerance mechanisms outside the thymus) is required to keep T-cells from reacting to these TRAs, although it was hard to explain how this is estab-lished. This is not necessary anymore with the recent discovery of a mechanism termed promiscuous gene expression (pGE) [107, 108]. This mechanism allows mTECs to ex-press antigens from all organs and tissues and even from developmentally and temporally regulated genes [54, 75, 181]. With this discovery several questions arose.

There were two competing hypothesis in which way these TRAs are expressed. Either randomly, that is an mTEC expresses antigens from different tissues at the same time, or in a tissue emulating pattern, that is an mTEC plays the role of a cell from a specific tissue and only expresses antigens from this tissue. The latter hypothesis suggests that in different compartments of the medulla different tissues are emulated and T-cells learn there to be tolerant to the specific antigens. This hypothesis is rather intuitive as this should eliminate T-cells quite efficiently and the mechanisms for gene expression are copies of the actual mechanisms in the tissue cells. However, the reality looks different. TRAs are expressed randomly. There is no indication that antigens from specific tissues are expressed more often together [116]. Instead, mTECs express random TRAs co-localised in chromosomal clusters [75, 102].

A second question deals with the regulation of pGE. The central and until now the only known molecule in this context is the autoimmune regulator (AIRE) [6], although some co-regulators like the interferon pathway begin to appear [72]. The discovery of TRAs which can be expressed independently of AIRE implies LTβR to regulate their expression [174, 54]. However, the outstanding role of AIRE in pGE regulation could be shown in several experiments with AIRE deficient or altered mice, that lead to different autoimmune diseases [6, 83, 159]. Interestingly, pGE is highly conserved between mouse and human [75], which underlines its importance in the course of evolution and is an explanation for many similar autoimmune diseases in both vertebrates. It is evident that AIRE regulates TRA expression directly and via epigenetic mechanisms, but its exact

(29)

function(s) are not identified and several models exist [199].

To understand pGE it is also of interest to have a look at the mTECs themselves. They share the same progenitor cells as the cTECs, but little is known about the first lineage decisions to becoming mTECs [18]. Medullary thymic epithelial cells can be divided into different subsets depending on the expression of the surface markers CD80 , CD40, MHC class II and AIRE. It has recently been shown that mTEC differentiation follows the so called ’terminal differentiation model’, that is mTECs develop from CD80low AIRE− (few CD80 molecules, no AIRE) to CD80high _AIRE+ _{(many CD80 molecules,}

AIRE present) and meanwhile expand the repertoire of genes they express [116]. It is suggested that there exists a unique mTEC lineage up to the mature mTECs [81]. The fully mature mTECs with the widest ranges of gene expression only survive about 2 weeks, which could be due to an overload of the gene expression machinery [114]. The mTECs in the medulla form separated areas called microdomains, which consist of one to three clonal islets [199, 18]. These islets again contain varying numbers of mTECs of all developmental stages.

One problem with mTECs is their limited capability of inducing T-cell apoptosis and their short lifetime in which they can present TRAs [68]. DCs on the other hand are highly capable of T-cell apoptosis induction and many of them are present in the medulla. It was shown that some of these DCs derive TRAs via so-called cross-presentation from mTECs. How this acquisition is achieved is under discussion, one mechanism might just be the collection of cell material from mTECs that died by apoptosis [76, 77]. This could help to magnify the effectivity of negative selection to TRAs. Furthermore, a special set of DCs together with a special set of mTECs is implied to be the mediator of Treg development in the medulla [217].

In a nutshell, pGE provides for an expression of tissue restricted antigens, whereby every antigen is expressed by only some of the mTECs, randomly in spatially distinct regions of the medulla. Only a certain number of TRAs is expressed simultaneously by neighboring mTECs [199]. Somehow there has to be a mechanism to ensure the effectiveness of pGE in order to prevent autoimmune diseases, especially in light of the fact that in some situations the loss of expression of only one particular TRA can have devastating effects [57]. This mechanism is implied to be stochastic and has to involve epigenetic regulators that are responsible for chromosomal remodelling [55, 215]. It remains to be shown how it really works.

Mature T-cells which survive negative selection eventually migrate to the periphery. The exact mechanism behind this is not identified, but it appears to happen in an ordered fashion, that is the oldest T-cells leave first [129].

2.2.4 T-Lymphopoiesis in numbers

After the qualitative description of T-cell development, we here describe it briefly quan-titatively.

On average about 10 − 100 hematopoietic precursors enter the thymus per day [116]. But, as already mentioned, this happens in a cyclic manner, such that one week after

(30)

leaving the bone marrow these cells enter the thymus and start seeding [60]. It then follows a stage of proliferative expansion and further differentiation over 10 − 14 days [154]. In total, T-cell production has a periodicity of about 3 − 5 weeks [66]. Upon T-cell linage commitment a T-cell divides about 20 times, mostly in the 2 week long double negative stage [116, 16]. All in all the thymus produces about 5 · 107 _{T-cells daily, but}

only 1 − 2 · 106 mature T-cells are released.

Upon β selection a T-cell has to be positively selected. The T-cell constantly tries to bind to pMHC molecules on DCs and cTECs, meanwhile editing the α chain of its receptor via sequential recombination rounds. This process lasts about 3 − 5 days, resulting in the death of the T-cell, if it is not positively selected. The theoretical repertoire size of unique TCRs which can thereby be created is > 1015 _{for mice and}

> 1018 for humans [46, 213]. About 3% of all T-cells survive thymic selection (10% survive positive selection from which again only 35% survive negative selection, other estimates are up to 60% survival rate during negative selection), which reduces the number of possible unique TCRs in the periphery to about 1013 for mice and 1016 for humans [175, 128, 56]. However, the estimated number of TCRs in the periphery is much lower (108 _{mice, 10}12 _{humans) and the number of unique TCRs is only 10}6 _{for mice and}

107 for humans [8, 30, 145]. This observation and the fact that there is a pool of TCRs, called public TCRs, which are present it many humans, lead to the theory of convergent recombination. It could be shown that certain TCRs can in principle be generated with a much higher probability [214, 213]. The overall effect of convergent recombination is until now not quantified and therefore it is not clear if it suffices to explain the quite small size of the unique TCR repertoire.

On the other hand we have the vast amount of possible antigens. This number is theoretically estimated to be in the same range as the original number for the TCRs, that is 1017 _{or higher [127]. But in every individual this number is much lower. Here,}

the restriction is imposed by the MHC molecules, as they cannot bind to all antigens. On the contrary, it is estimated theoretically and experimentally that every type of MHC can ’only’ bind to about 109 _{different antigens [67]. This also underlines the}

need for several different MHC molecules in order to expand the space of presentable antigens. However, this number is still higher than the number of unique TCRs in an individual. Consequentially, a T-cell has to be cross-reactive (also termed poly-specific). Experimental measurements imply that a TCR can bind to about 106 antigen types [67]. Generally, these types share similarities on the molecular level, that is a T-cell cannot bind to different totally distinct antigens. This suggests that negative selection creates ’holes’ in the space of detectable antigens, which could be shown by an experiment where a antigens resulting from HIV share similarities with self antigens and thereby evade detection [67].

It can be shown that prior to entering the selecting stromal environment T-cells move with an average speed of 3 − 8µm/min and top speed of 30µm/min, following random trajectories [16]. This changes during (positive) selection. The T-cells scan APCs and bind to them. Two different contact types with cTECs were identified, a short one lasting 13−36 minutes and a long one lasting over 6 hours. Upon positive selection T-cells move more rapidly and in a more directed fashion in direction of the medulla. In the medulla